Model comparison
Gemma 4 31B IT vs Inkling-Small
Inkling-Small is the stronger model overall, scoring 46.5 to 43.5 on the Noometry Index. Gemma 4 31B IT costs 4.2× less per token, which makes it the better buy when Inkling-Small's lead doesn't matter for your workload.
Last verified . 24 shared benchmarks.
GoogleSummary
- They share 24 benchmarks with published results for both. Gemma 4 31B IT scores higher in 5 categories and Inkling-Small in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Inkling-Small leads 38.6 to 27.2.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 73.3% for Gemma 4 31B IT and 90% for Inkling-Small.
- Gemma 4 31B IT is cheaper at $0.09 / $0.34 per million input/output tokens, against $0.45 / $1.20 for Inkling-Small.
- Inkling-Small accepts more context: 524K tokens versus 262K.
Side by side
| Gemma 4 31B IT | Inkling-Small | |
|---|---|---|
| Provider | Thinking Machines Lab | |
| Noometry Index | 43.5 | 46.5 |
| Released | 2026-04-02 | 2026-07-15 |
| Weights | Open | Open |
| Context window | 262K | 524K |
| Max output | 33K | 1.05M |
| Input $ / M tokens | $0.09 | $0.45 |
| Output $ / M tokens | $0.34 | $1.20 |
| Results tracked | 35 | 33 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Inkling-Small leads
Gemma 4 31B IT: 42.3 (#108), Inkling-Small: 43.6 (#85)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| LMArena WebDev | 1366 | 1409 |
| SciCode | 43.4% | 48.7% |
| LMArena Coding | 1459 | 1451 |
| WeirdML | 52.3% | — |
| ALE-Bench | 925.5 | — |
Reasoning Inkling-Small leads
Gemma 4 31B IT: 27.2 (#122), Inkling-Small: 38.6 (#63)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| CritPt | 1.4% | 8.3% |
| Chess Puzzles | 5% | 18% |
| LMArena Hard Prompts | 1448 | 1423 |
| Epoch Capabilities Index | 142.74 | 150.15 |
| ARC-AGI-2 | — | 40.1% |
| Kagi LLM Benchmark | 63.5% | — |
| NYT Connections (extended) | 70.6% | — |
| ARC-AGI-1 | — | 84% |
| Thematic Generalization | 53% | — |
| Mystery Game Puzzles | — | 6% |
| DTBench | 82.7% | — |
| LMCA | 39.3% | — |
| Surface Evolver Bench | 30.6% | — |
Math Inkling-Small leads
Gemma 4 31B IT: 43.2 (#81), Inkling-Small: 45.1 (#77)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.3% | 90% |
| LMArena Math | 1465 | 1459 |
| FrontierMath (Tiers 1-3) | — | 46.3% |
| FrontierMath Tier 4 | — | 17.1% |
| ProofBench | — | 6% |
Knowledge Inkling-Small leads
Gemma 4 31B IT: 37.9 (#151), Inkling-Small: 48.2 (#77)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| GPQA Diamond | 75.8% | 88.5% |
| SimpleQA Verified | 10.4% | 19.1% |
| LMArena Expert | 1465 | 1442 |
| Vectara Hallucination Rate | 7.4% | — |
Multimodal Gemma 4 31B IT leads
Gemma 4 31B IT: 41.6 (#34), Inkling-Small: 39.1 (#62)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| LMArena Vision | 1277 | 1235 |
| LMArena Document | 1425 | — |
Multilingual Gemma 4 31B IT leads
Gemma 4 31B IT: 53.8 (#57), Inkling-Small: 51.7 (#104)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| LMArena Non-English | 1431 | 1402 |
| LMArena Chinese | 1476 | 1465 |
| LMArena French | 1435 | 1436 |
| LMArena Russian | 1460 | 1391 |
| LMArena Spanish | 1444 | 1428 |
| LMArena German | — | 1405 |
| LMArena Japanese | — | 1405 |
| LMArena Korean | — | 1363 |
Instruction Following Gemma 4 31B IT leads
Gemma 4 31B IT: 75.5 (#61), Inkling-Small: 73.8 (#114)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| LMArena Instruction Following | 1433 | 1399 |
Long Context Gemma 4 31B IT leads
Gemma 4 31B IT: 44.2 (#71), Inkling-Small: 42.7 (#118)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| LMArena Longer Query | 1446 | 1401 |
Writing & Preference Too close to call
Gemma 4 31B IT: 60.5 (#96), Inkling-Small: 59.6 (#107)
| Benchmark | Gemma 4 31B IT | Inkling-Small |
|---|---|---|
| LMArena Text | 1443 | 1414 |
| LMArena Creative Writing | 1415 | 1331 |
| EQ-Bench Creative Writing | 1368 | 1491 |
| LMArena Multi-Turn | 1452 | 1418 |
| EQ-Bench 4 | 1120 | — |
Frequently asked questions
Is Gemma 4 31B IT better than Inkling-Small?
Inkling-Small is the stronger model overall, scoring 46.5 to 43.5 on the Noometry Index. Gemma 4 31B IT costs 4.2× less per token, which makes it the better buy when Inkling-Small's lead doesn't matter for your workload.
Which is cheaper, Gemma 4 31B IT or Inkling-Small?
Gemma 4 31B IT is cheaper. It lists at $0.09 per million input tokens and $0.34 per million output tokens; Inkling-Small lists at $0.45 and $1.20.
Is Gemma 4 31B IT or Inkling-Small better for coding?
Inkling-Small scores higher on coding benchmarks: 43.6 versus 42.3 in the Noometry coding category.
Which has the bigger context window?
Inkling-Small does, with 524K tokens against 262K.
How many benchmarks do Gemma 4 31B IT and Inkling-Small share?
24 benchmarks have published results for both models. Gemma 4 31B IT has 35 scored results on Noometry and Inkling-Small has 33.